CORRELATION ANALYSIS BETWEEN PARK LANDSCAPE AND URBAN THERMAL ENVIRONMENT BASED ON IMPROVED MSRN2 ALGORITHM

With the acceleration of urbanization and the intensification of urban heat island effect, traditional remote sensing methods have the problem of insufficient multi-scale feature capture when finely analyzing the connection between thermal environment and landscape elements.To address this issue, an improved multi-scale residual network 2 algorithm is proposed, which introduces a texture color dual branch feature extraction module and attention mechanism to optimize the dynamic fusion ability of multisource remote sensing data, and combines the cellular Markov model to predict the spatiotemporal evolution of urban thermal environment, to explore the correlation between park landscape and thermal environment.The experimental results show that the improved multi-scale residual network 2 algorithm has a feature extraction accuracy of up to 0.98, with an efficiency improvement of about 60% compared to traditional models, which is a significant enhancement.In addition, the overall error of the cellular Markov model for thermal environment prediction is less than 5%.Correlation analysis shows that the degree of fragmentation in park landscapes is significantly correlated with the expansion of high-temperature areas, and the heat island effect exhibits a growing tendency.The research verifies the effectiveness of the improved algorithm and prediction model, providing scientific basis and technical support for optimizing urban green space planning and alleviating thermal environment problems.